Urban sound tagging with spatiotemporal context

Abstract

Sound classification is useful for city planner to understand the noise profiles of urban environments such as traffic noise, construction noise, and community noise in order to mitigate urban noise pollution. Metadata such as when and where a sound event occurs are potentially useful for urban sound classification. In this project, you are going to investigate the usages of metadata such as timestamps and locations of the sound events in order to improve the accuracy of the sound classification system in urban environment setting.